Giampaolo Tortora

Biomarcatori di Sensibilità e Resistenza nei Tumori del Rene

Cattedra di Oncologia Medica UOC Oncologia Medica

Comprehensive Cancer CenterScuola di Medicina e Chirurgia e

Azienda Ospedaliera Universitaria IntegrataVerona

Nature Rev Clincal Oncology, 2013

VHL complexdisrupted

Multiproteincomplex

Endothelial stabilisation

Autocrine growthstimulationAngiogenesisOrgan-specific

metastasis

Bernards R. Nature 2003;425:247–8Lineham WM, et al. J Urol 2003;170:2163–72

Consequences of VHL gene mutation

VHL = von Hippel–Lindau; HIF = hypoxia-inducible factorCXCR4 = chemokine receptor 4

VEGF TGF-αPDGFCXCR4

HIF1-, HIF2-accumulation

HIF HIF HIF

VHL proteinMutant

α-domain-domain

Adapted from M. Hall

HIF1 è coinvolto in tutti gli RCC

Rheb

TSC2TSC1

Amino acidsAmino acids

LKB1AMPK

FKBP-rapamicina

mTOR

Recettore dei fattori di crescita

IRS

PI3KPDK1Akt/PKB

Growth Growth factors factors

(insulin/IG1(insulin/IG1

PTEN

HIF

BHDCromofobooncocitoma

Ipossia

Perdita di PTEN

METPapillare di tipo 1

TSCCellule chiare

VHLCellule chiare

FHPapillare di tipo 2

AmminoacidiFattori di crescita

(insulina/IGF1)

HIF1-HIF1- Degradazione di HIF

Degradazione di HIFNormossiaNormossia

VEGFVEGF

Ipossia o perdita di VHL

Ipossia o perdita di VHL

VHLVHL

UbiquitinazioneUbiquitinazione

Mancata ubiquitinazione e inattivazione di HIF1

Mancata ubiquitinazione e inattivazione di HIF1

mTOR/TSC2mTOR/TSC2

mTOR integra i segnali di nutrienti e fattori di crescita

mTOR

Glucosio

ATPAminoacidi

TSC2

TSC1

AMPK

Glucosio

ATP

Segnale di crescita

PI3K

Akt

LAT

GLUT 1

mTOR è un sensore per la disponibilitàdi amminoacidi, rifornimenti metabolici ed energia I depositi di nutrienti e di energia sono

essenziali per la sintesi proteica, la crescita cellulare e la sopravvivenza. L’ attivazione di mTOR può aumentare

l’espressione dei trasportatori dei nutrienti, Aumentando l’accesso cellulare ai

rifornimenti metabolici, mTOR può sostenere la crescita cellulare

Sintesi proteica

Angiogenesi

BioenergiaProliferazione cellulare

mTOR Signaling pathways

mTOR complexes

Akt

PI3KPTEN

Growth factors

4E-BP1p70S6K1elF-4E

mTORC1

Cell survivalApoptosis

mTORC2

FOXO3A BAD

TSC2

TSC1

EverolimusTemsirolimus

Resistance

Proliferation

mTORC is linked to HIF-1 and angiogenesis

Tijeras-Raballand et al. Targ Oncol (2012) 7:173–181.

Other proangiogenic factors and systems relevant in RCC pathogenesis and resistance to treatment

• IL-8• bFGF and FGFRs• HGF/c-MET

• SDF1/ CXCR4

Effects of drug sequences on growth, signal transduction and cytokine expression of 786-O tumors xenografted in nude mice

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 180

20

40

60

80

100

120

ControlSunitinib chronicSunitinib>SorafenibSunitinib>Everolimus

Weeks

Perc

ent o

f sur

vivi

ng m

ice

pAkt

Akt

pp70S6K

p70S6K

pMAPK

MAPK

actin

HIF-1

VEGF

Sunitinib ch

ron

Sunit > Soraf

enib

Sunit > Eve

rolim

us

Rosa R et al., British J Cancer, 2013

The role of HIF-1 as a critical regulator of BMDC in tumor

SDF-1 (CXCL12) is increased in hypoxic areas and stimulates angiogenesis and metastasis

Staller P. et al. Nature 425: 307, 2003

CXCR4 expression in human renal cell carcinoma

Red bars, CXCR4; blue bars, GLUT1; green bars, CA9.

TUMOR

Controlateral normal kidney

Zagzag, D. et al. Cancer Res 2005 ;65:6178-6188

Epigenetic expansion of VHL-HIF signal output drives multiorgan metastasis in RCC

Effects of VHL reintroduction on CXCR4 expression measured by quantitative RT-PCR analysis. Expression normalized to 786-O cells transduced with the control vector

Kaplan-Meier analysis of metastasis-free survival in patients with ccRCC. Classification based on CXCR4 mRNA expression in primary ccRCCs, top 1/3 shown in red, bottom 2/3 in blue.

VHL-HIF Signal Output Modulation Associated with ccRCC Progression

Hypothesis that when VHL loss stabilizes HIF, early ccRCC genes such as VEGF and ADM are highly expressed whereas the activation of prometastatic HIF target genes such as CXCR4 and others requires additional amplifying mechanisms.

Vanharanta S et al…Massague J., Nature Medicine 2013

Methylation and inactivation of genes is common in ccRCC

Mahers ER, Seminars in Cancer Biology 23: 10-17, 2013

Methylation and inactivation of:•VHL in 10-30% of cases in sporadic ccRCC •Ras association domain family 1 gene (cell cycle negative regulator) in 45% of cases•Wnt antagonist Frizzled related protein 1 in 35-68% •Gremlin1 correlates with a poorer OS

Biomolecular markers under investigationDescription Sample/data 

collection method Biomarkers under evaluation

VHL and HIF deletion or inactivation 

Collection and assay of tumour biopsy (frameshift, nonsense, splice and in‐frame deletions/insertions)

No relationship was found between VHL deletion or inactivation and outcome of RCC patients treated with either TKIs or mTOR inhibitors (Sonpavde and Choueiri).

HIF expression Frozen tumour HIF levels (by western blot)

Increased HIF‐1a and HIF‐2a expression may predict response to sunitinib (Patel 2008)

Carbonic anhydrase IX (CAIX)

Collection and assay of tumour biopsy samples 

A potential predictive role of CAIX was not demonstrated with Il2 (McDermott 2010). 

No correlation between CAIX and outcome after TKIs and mTOR inhibitors (Sonpavde and Choueiri). 

VEGF and VEGF‐Rs  Collection and assay of serum samples at regular intervals

Collection of tumour  tissue..Serum samples (sVEGFR2) plasma, at regular intervals. 

Correlation found between lower baseline levels of VEGF and response to sunitinib and between lower baseline levels of soluble (s) VEGFR‐3 or its ligand VEGF‐C and longer PFS to sunitinib (Rini 2008). 

Decrease of sVEGFR‐2 correlates with PFS and tumor response to pazopanib (Hutson 2008).

Nature Rev Clincal Oncology, 2013

Biomolecular markers under investigationDescription Sample/data 

collection method Biomarkers under evaluation

Cytokines and angiogenic factors

Collection of plasma  samples at regular intervals Concentrations of plasma cytokine and angiogenic factors (CAFs)

IL‐6, IL‐8, VEGF, HGF, Osteopontin and TIMP‐1 correlated with outcome in patients receiving pazopanib. In particular, in the phase III trial IL‐6, IL‐8, HGF and Osteopontin correlated with PFS (Tran et al. 2012). 

A signature was identified using six biomarkers (sCAIX, Osteopontin, VEGF, TRAIL, Collagen V, and sVEGFR‐2) with a strong correlation with PFS in sorafenib treated patients (Zurita 2012).

Single nucleotide polymorphisms (SNPs). 

Collection and assay of samples at regular intervals. (Germline DNA extracted from peripheral blood)

Combined presence of SNPs 889 and 1416 correlated with improved outcome in patients treated with sunitinib (Kim 2012). 

SNPs in the IL‐8, FGFR2, VEGF‐A and VEGFR‐3 associated with Overall Survival in pazopanib treated patients (Xu 2011). 

Relationship between SNPs in the VEGF pathway genes and PFS and blood pressure in Axitinib treated patients (Rini2012).           

Nature Rev Clincal Oncology, 2013

Prognostic and predictive Biomarkers in RCC

• Immunohistohemistry• Circulating in plasma or serum • Polymorphisms (SNPs)

Clin Cancer Res; 20(8) April 15, 2014

Circulating biomarkers

Markers of resistance to Tivozanib

Hutson T... Choueiry T, ESMO 2014

• Low myeloid index is associated with significantly longer PFS • Longer PFS based on median cutoff (PFS 14.7 vs 8.3 months, hazard ratio [HR] 0.49,

P=0.035; 95% CI 0.25–0.96) and as a continuous variable (P=0.03; N=63)• The gene signature did not exhibit a prognostic effect in a historical datase

Genetic Variations in Angiogenesis-Related Factors

1. Kim JJ, Vaziri SA, Elson P, et al. J Clin Oncol. 2010; 28:15 (suppl):abstr 4629. 2. Xu C et al. J Clin Oncol. 2011;29:7(suppl):abstr 303.3. Ball HA et al. J Clin Oncol. 2010;28(suppl 15):abstr 4520. 4. Eechoute K et al. J Clin Oncol. 2010;28(suppl 15):abstr 4521.

SNPs as Predictive Factors in RCC

Association of IL8 polymorphisms with OS in patients with RCC inCOMPARZ (pazopanib versus sunitinib phase III study).

• 724/110 pts analyzed (65%) (pazopanib, N = 371; sunitinib, N = 353).• Associations of IL8 polymorphisms (rs1126647 and rs4073) with PFS and OS tested using

the Cox proportional hazards model in a combined analysis of all patients and in pazopanib-and sunitinib-treated patients. One-tailed P values were calculated for effects in the same direction.

• PFS: No significant association in the combined analysis or in pazopanib-treated patients, significant association only in sunitinib-treated patients (P = 0.017).

• OS : significant associations in the combined analysis (P = 0.010) and in sunitinib-treatedpatients (P = 0.0043). Not in pazopanib-treated patients (P = 0.30).

• HRs for genetic effects were not significantly different between sunitinib- and pazopanib-treated patients (two-tailed P = 0.23 for genotype-by-treatment interaction).

• Kaplan-Meier plots suggested a recessive genetic model in the combined data set, with median OS (95% CI) 23.7 months (15.4–29.1) for rs1126647 TT genotype compared to 35.5 months (30.8–∞) for AA or AT genotypes (HR = 1.66, P = 0.0007). Similar associations seen for rs4073.

• Conclusion: Germline variants in IL8 are associated with survival outcome in patients with RCC who have received angiogenesis inhibitors.

Xu CF et al, ASCO 2013, J Clin Oncol 31, 2013 (suppl; abstr 4519)

Many potential prognostic and predictive molecular biomarkers have now been identified in RCC,

although none has yet entered into clinical practice, and all require prospective validation in appropriately designed

randomized studies

Clin Cancer Res; 20(8) April 15, 2014

Different mechanisms for intrinsic vs. acquired resistance

J.C. van der Mijn et al. Drug Resistance Updates (2015)

Tumors with low contrast enhancement, high glucose uptake (PET) appear intrinsically resistant to VEGF targeted therapies

Acquired resistance is associated with rebound vascularization, potentially mediated by IL-8, HGF/MET, VEGFR3 and/or FGFR2.

Lysosomal sequestration of hydrophobic weakly basic drugs might be a resistance mechanism that entraps these cytotoxic agents away from their intracellular target

Predictive biomarkers with statistical significant results.

T. Funakoshi et al. / Cancer Treatment Reviews 40 (2014) 533–547

Istotipi e kinoma nel carcinoma renale

Linehan, W. M. et al. Clin Cancer Res 2007;13:671s-679s

Clear CellClear Cell Chromo/OncoChromo/Onco Papillary I/IIPapillary I/II WTWT

Kinome expression (74/518 kinases with >3-fold differential expression vs.normal kidney (Teh, ASCO 2007)

Gerlinger M et al. N Engl J Med. 2012;366:883-892.

• 55% of all somatic mutations (70/128) from the primary tumor were detected in a single biopsy sample

• When multiple samples from the same tumor were taken from different regions, only 34% of mutations were detected in all samples– Number falls to 31% if pre-treatment and metastases are included

Gray, presence of mutation.Dark blue, absence of mutation.

Gerlinger et al. Nat Genet. 2014 Mar;46(3):225-33

• Ultra-deep sequencing (multiregion exome sequencing - M-seq) identified Intratumor heterogeneity in all cases.

• 75% of identified ccRCC driver aberrations are subclonal, confounding estimates of driver mutation prevalence.

• ITH increased with number of biopsies analyzed, without evidence of saturation in most tumors.

• Chromosome 3p loss and VHL aberrations were the only ubiquitous events

Gerlinger et al. Nat Genet. 2014 Mar;46(3):225-33

Comparison of driver mutation prevalence in ccRCC samples

PBMR1 and BAP1

SW1/SNF chromatin remodelling complex gene polybromo1 (PBRM1)

BRCA related protein-1 (BAP1) is inactivated in 15% of ccRCCs

PBRM1 and BAP1 were largely observed to occur mutually exclusiveIn comparison with the PBRM1 mutation, BAP1-deficient tumours were of higher grade and had distinct gene expression profiles

Loss-of-function mutations in the VHL allele are thought to represent an early event in ccRCC development not sufficient alone to drive tumour growth.

Recurrent ccRCC gene mutations and their impact on prognosis

Nature Rev Clincal Oncology, 2013

Kapur P et al., Lancet Oncol 2013; 14: 159–67

BAP1 and PBRM1 are prognostic factors for OS

Role of PBMR1 and BAP1 on Survival : Retrospective analysis fromtwo different datasets (UTSW and TCGA)

Sato Y. Et al., Nature Genetics 45: 860-869, 2013

• More than 100 ccRCC cases fully analyzed by wholegenome and/or whole-exome and RNA sequencing, array-based gene expression, copy number and/or methylation analyses.

• Identified a full spectrum of genetic lesions and analyzed gene expression and DNA methylation signatures and determined their impact on tumor behavior.

• Defective VHL-mediated proteolysis was a common feature of ccRCC, which wascaused not only by VHL inactivation but also by new hotspot TCEB1 mutations, which abolished Elongin C–VHL binding, leading to HIF accumulation.

• Other newly identified pathways and components recurrently mutated in ccRCC included PI3K-AKT-mTOR signaling, the KEAP1-NRF2-CUL3 apparatus, DNA methylation, p53-related pathways and mRNA processing.

• Unmasked new correlations between DNA methylation, gene mutation and/or gene expression and copy number profiles, enables the stratification of clinical risks for patients with ccRCC.

Nature Genetics 45: 849-850, 2013

• ccRCC is a disease of 3p loss

Cancer Genome Atlas Research Network, Nature 499: 43-49, 2013

Genetic changes underlying ccRCC include alterations in genes controlling cellular oxygensensing (i.e. VHL) and the maintenance of chromatin states (i.e., PBRM1). In more than 400

tumours studied using different genomic platforms, identified 19 significantly mutated genes.

• The PI3K/AKT pathway is recurrently mutated, suggesting it as as a therapeutic target. • Widespread DNA hypomethylation was associated with mutation of the H3K36

methyltransferase SETD2• integrative analysis suggested that mutations involving the SWI/SNF chromatin remodelling

complex (PBRM1, ARID1A, SMARCA4) could have far-reaching effects on other pathways. • Aggressive cancers demonstrated evidence of a metabolic shift, involving downregulation of

genes involved in the TCA cycle, decreased AMPK and PTEN protein levels, upregulation of the pentose phosphate pathway and the glutamine transporter genes, increased acetyl-CoA carboxylase protein, and altered promoter methylation of miR-21 (also known as MIR21) and GRB10.

• Remodelling cellular metabolism thus constitutes a recurrent pattern in ccRCC that correlates with tumour stage and severity offering new views on treatment opportunities.

DNA demethylation allows CYTIP expression driving metastatic ccRCC

Vanharanta S et al, Nature Medicine 2013

Changes in the epigenetic landscape allow hyperinduction of VHL-HIF target (i.e.: CXCR4 and CYTIP). If these genes are beneficial to the tumor they will be selected to drive ccRCC progression

metastasis-associated VHL-HIF target genes include cytohesin 1 interacting protein (CYTIP)and CXCR4.

Cancer Genome Atlas Research Network, Nature 499: 43-49, 2013 2013

Overall promoter DNA hypermethylation frequency in the tumour increases with rising stage and grade

DNA methylation in ccRCC

VHL oxygen and iron sensing

MET LKB1-AMPK energy sensing

FLCN binds AMPK and might interact with the cellular energy and nutrient sensing

TSC1 downstream of AMPK and negatively regulates mTOR in response toTSC2 cellular energy deficit

FH central role in the mitochondrial tricarboxylic acid cycle

SDH coupled to energy production through oxidative phosphorylation

Pathways responding to metabolic stress or nutrient stimulation

Molecular correlates of patient survival involve

metabolic pathways

• Worse survival correlates with upregulation of pentose phosphate pathway (G6PH, PGLS, TALDO and TKT), fatty acid synthesis (ACC and FASN), and PI3K enhancing (MIR21).

• Better survival correlates with upregulation of AMPK complex genes, multiple Krebs cycle genes and PI3K pathway inhibitors (PTEN, TSC2).

• Additionally, specific promoter methylation events, including hypermethylation of PI3K repressor GRB10, associate with outcome

Worse outcome Better outcome

Cancer Genome Atlas Research Network, Nature 499: 43-49, 2013 2013

Ingenuity Pathway Analysis of deregulated genes in ccRCC vs. normal tissue samples

Ingenuity Pathway Analysis revealed HIF1A, STAT1, STAT3, SP1 and LHX1among the top Transcription Factors of the top 1% deregulated genes in ccRCC vs. the normal tissue samples, among the 5 Oncomine datasets

The genes forming the top 5 gene networks as identified by IPA (score>25), participate in:

• Hematological and reproductive systems development and function, cell-to-cell signaling and interaction (score=35);

• Carbohydrate metabolism, cell death, endocrine system disorders (score=33);

• Lipid metabolism, small molecule biochemistry, molecular transport (score=26)

• Molecular transport, renal and urological disease, cellular function (score=28)

• Small molecule biochemistry (score=30);

Zaravinos A et al, Oncoscience. 2:117-31, 2014

Presented By Samuel Kaffenberger at 2015 Genitourinary Cancers Symposium

• Integrated genomic and clinical information on 324 patients with ccRCC• RPPA data: 166 proteins/phosphoproteins targeted toward specific pathways

• Promise of genomics in predicting drug sensitivity• Imperfect correlation between mutations/CNA/mRNA expression and phenotype• Proteomics: Convergence of DNA and RNA expression

Background

cTCGA Project

Results: 5-cluster heat map

Cluster heat map

Modified from Samuel Kaffenberger ASCO GU 2015

Modified from Samuel Kaffenberger ASCO GU 2015

Novel antibody panel for prediction of response to mTOR inhibitors

IHC of downstream mTOR pathway phosphoproteins : Extreme Responders

Novel antibody panel used in patients with extreme response to rapalogs (everolimus)